A multi-stage game model for the false data injection attack from attacker’s perspective

https://doi.org/10.1016/j.segan.2021.100541Get rights and content

Highlights

  • The false data, attack range and results are analyzed from attacker’s perspective.

  • A multi-stage game model is proposed based on two-person zero-sum game theory.

  • The strategic adjustments and multi-path attacks are emphasised in multi-stage game.

Abstract

With the rapid development and widespread application of information and communication technology, power system has gradually become a cyber physical power system (CPPS) in which the cyber space and physical space are deeply coupled. The continuous interaction between information flow and power flow renders power grid confronted with potential cyber attack risks. As a typical attack approach, false data injection attack (FDIA) interferes with control decisions making by weakening information integrity. In this paper, regarding the phasor measurement unit (PMU) as the attack-defense target in power system, a multi-stage game model for FDIA is proposed based on attacker’s perspective, putting particular emphasis on data tampering, strategic adjustments, and multi-path attacks in a multi-stage game. Firstly, from the perspective of the attacker, the false data is constructed based on the PMU configuration, the attack range is optimized, and the attack consequences are quantified. Secondly, based on two-player zero-sum game theory, take the overall revenue and multi-path attacks in multi-stage game into consideration, the optimal attack-defense strategy is obtained by solving the Nash equilibrium point. Finally, experiment results of multi-stage game and one-stage game are discussed. The simulation results demonstrate that the proposed multi-stage game strategy owns better effect and applicability for attackers.

Introduction

With great advances of sensors, computers, and communication networks, traditional power system has been transformed into a multi-dimensional complex system integrating computation, communication, and control, which is universally known as a cyber physical power system (CPPS) [1], [2], [3], [4]. On the one hand, it promotes the ability of state perception, real-time analysis, scientific decision-making, and precise execution in power resource; on the other hand, much vulnerability in communication networks and information equipment may be utilized by malicious entities, bringing about potential network risks [5], [6]. Different from the traditional Internet, there is a close coupling relationship between the power communication network and the control operation of the physical power grid. Once invaded, it will cause chain damage to the power grid, leading to a large-scale blackout, which not only causes huge economic losses, but also affects the normal life of residents [7]. In recent years, a series of cyber-security incidents in CPPS have occurred at home and abroad, causing significant losses in worldwide, such as the Stuxnet virus attack against Iran nuclear power plant in 2010 [8]. Stuxnet is the first industrial control system virus, which was specially developed to attack the Siemens control system in Iran’s nuclear facilities, resulting in about 30,000 network terminals were infected. As another common example, the Black Energy virus incident against the Ukraine power grid in 2015 [9], which brought about power cut lasting 6 h for 1.4 million users. Besides, the network attack against Venezuela hydropower stations in 2019, which led to blackout in 16 states across the country. In addition, as is reported, provincial power utilities are attacked premeditatedly more than 2000 times a month in China [2]. Due to the threat and destructiveness in cyber-attack, the problem of CPPS cyber security cannot be ignored.

According to the purpose and influence extent of cyber attack in power system, it can be divided into three categories: destroying availability, integrity and confidentiality of information [10]. False data injection attack (FDIA) is a common way to destroy the integrity. It means that the attacker has the capability to bypass the bad data detection mechanism, inject false measurement data through intruding instruments and sensors, manipulate the state estimation arbitrarily, thereby interfering with the normal operation of the power system [11], [12], [13], [14]. In addition to attacking static estimation, FDIA can also be used to attack other important modules of smart grids, such as electricity prices, distributed energy status, and dynamic partition of microgrids [15]. In terms of attackers, a well-designed FDIA will induce the control center to misjudge the current state of the power grid, fail response or nonresponse occurrences on control device, even pose a serious threat to the overall security of the power grid. Therefore, FDIA is an effective attack method with sufficient capability.

Recently, scholars have conducted researches on CPPS attack and defense strategies. In the aspect of attack, researchers mainly study the impact of FDIA behavior, involving electricity market, power system operation, and distributed energy routing. In [16], the paper designed a FDIA based on the stochastical model against the state estimation of power market, which leads to financial misconduct; in [17], when the attacker accurately grasps the topology structure, system parameters, state estimation algorithm and bad data detection method, an effective FDIA will seriously damage the stability of power system; in [18], forged energy information or link state information can be injected into the energy request and response messages between nodes, resulting in an imbalance between power supply and demand, even worse, disturbing the energy distribution. Based on that, the defense process against cyber attack can be classified into two parts: detection and protection [19]. In terms of detection, the information from cyber side, physical side, and bilateral detection is utilized to discover the behavior of the system suffering from FDIA in time [20], [21], [22]. In terms of protection, to ensure the information security, authentication verification, encryption, and network isolation are taken as effective measures for resisting attacks and intrusion actions [23], [24]. Apart from that, the defender can adopt the moving target defense strategy to change the network topology as well, so that the attacker is incompetent in mastering the accurate system parameters, to provide guarantee for the safe and stable operation of the physical side [25].

Some achievements have been made in unilateral researches, while in order to achieve better effect in attack and defense, the interaction between the attacker and the defender needs to be considered. There is no doubt that the essence for cyber attack and defense can be considered as dynamic changes of mutual influence between the bilateral strategies, making reasonable decisions with limited resources. The antagonism, strategy dependence and non-cooperation are the basic characteristics of game theory.

Regarding the problem of a joint game, the majority of researches have focused on the aim of stability reduction in power system and economic loss in electricity market. In view of stability reduction in power system, a stochastic zero-sum game theory was utilized to simulate the attack against lines in [26], quantifying the attack consequence by the optimal load shedding algorithm. The optimal defense decision was obtained under Nash equilibrium point, which protected the power grid from attackers. In the same way, in [27], with data fusion, the combination of Markov decision process and game theory was studied in network security situation, the process of attack and defense game determined the rules for network security state transition, to provide technical support for network security defense. In view of economic loss in electricity market, a game theory based on optimal time-of-use electricity price strategy was proposed in [28], designing utility function and reverse induction method for the purpose of getting Nash equilibrium solution, the results demonstrate it improved the efficiency of the electricity market significantly.

The traditional game models mentioned above have imperfections, previous researchers took advantage of rational assumption aiming to the optimal strategy through a game completely, without regarding the process of strategy adjustment for two players in different stages. Considering the multi-stage and continuous characteristics of attack-defense confrontation, it is more reasonable to be considered as a multi-stage process. In addition, the studies only devoted mind to how to allocate resources and optimize the interaction between the two players, while simplifying the power service details, such as data tampering, strategic adjustments, and attack paths, which reduces the effectiveness of model and method mathematically.

On the foundation of summarizing and analyzing previous research results, the paper puts forward a multi-stage game model for the FDIA based on attacker’s perspective. To be more specific, the attacker intrudes and tampers with PMU measurements to bring about load shedding of the system, while the defender detects the attack behavior by means of deploying redundant PMU devices. For this reason, the attack-defense game space is composed of the two actions. Based on the two-player zero-sum game theory, attack behavior of the PMU is simulated in UIUC 150-bus system, and the optimal attack-defense strategy in single stage is derived by solving the Nash equilibrium point. In addition, as mentioned, the paper lays great emphasis on a multi-stage game method. The simulation results show that it is effective for the attacker to consider the overall benefits of multi-stage and launch multiple attacks in a short time compared with the one-stage game, which verifies the validity and applicability of the proposed model. The main contributions in this paper are given as follows:

(1) The construction of false data, the optimization of attack range as well as PMU configuration, and the quantification of attack consequences are analyzed from the attacker’s perspective.

(2) The optimal attack-defense strategy in one-stage and multi-stage game is obtained by solving the Nash equilibrium point respectively, which is based on two-person zero-sum game theory.

(3) Compared with the one-stage game, the data tampering, strategic adjustments, and multi-path attacks are put particular emphasis in multi-stage game.

The following chapters are organized as follows: Section 2 introduces the FDIA model. In Section 3, the optimal PMU configuration, the optimal attack range, and the optimal load shedding model are expounded in detail from the perspective of the attacker. The one-stage and multi-stage game models based on the two-player zero-sum game theory are presented in Section 4. Section 5 gives the simulation results of one-stage and multi-stage game in UIUC 150-bus system and the results are analyzed and discussed. Finally, the paper is concluded in Section 6.

Section snippets

Background and taxonomy of FDIA

An integrated, high-speed, and bidirectional communication system enables the real-time interaction between information system and electric power system in smart grid, as shown in Fig. 1. In the smart grid, all measurement data collected from remote terminal units (RTUs) and PMUs are transmitted to supervisory control and data acquisition (SCADA) system through network [29]. Further, the gathered data in SCADA are communicated to the energy management system (EMS) for implementing state

Optimal configuration of PMUs

The observability of the entire power system means that the voltage phasor of each node in the network can be observed directly by PMUs or indirectly according to Kirchhoff’s law and Ohm’s law. In this paper, the PMUs are considered as the measurement units in the system, carrying out an attack in the controllability and observability respectively. Owning to the high cost of installing PMUs in each node, a 0–1 integer programming model was established​ for optimizing the configuration of PMU

Model and solution of one-stage attack-defense game

In this paper, the attack-defense game research carried on is on the foundation of taking PMUs as the objective. From the attacker’s point of view, the attacker has the capacity to make an incursion and tamper with the measured value of the PMU devices abruptly. Nevertheless, from the perspective of defender, a certain quantity of redundant PMU devices ought to be deployed, which is used for detecting the attack behavior effectively. The combination of the two actions constitutes the game

Example setting

In this paper, the UIUC 150-bus system is taken as an example to simulate the attack-defense game process of FDIA. The UIUC 150-bus power system case contain substation geographic coordinates and other parameters useful for geomagnetic disturbance studies. The case is entirely synthetic, which is created from public information and a statistical analysis of real power systems, showing similar load and generation distributions in the location. The UIUC 150-bus topology is shown in Fig. 4. In

Conclusion

On the foundation of summarizing and analyzing previous research results, the paper puts forward a multi-stage game model for the FDIA from attacker’s perspective. The method takes the measurement unit PMUs as the attack-defense target, lays emphasis on the data tampering, attack-defense strategy adjustment, and multi-path attacks in the multi-stage game. In the planning stage, the attack mode is to tamper with the measurements of PMUs, and the defend means is to deploy redundant PMU

CRediT authorship contribution statement

Na Yi: Methodology, Software, Formal analysis, Writing – original draft, Investigation. Qi Wang: Conceptualization, Methodology, Validation, Formal analysis, Writing – original draft. Limei Yan: Resources, Writing – review & editing, Data curation. Yi Tang: Validation, Investigation, Data curation. Jianjun Xu: Funding acquisition, Resources, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

This work was supported by the National Natural Science Foundation Project of China (51774088).

References (29)

  • WangQ. et al.

    A two-layer game theoretical attack-defense model for a false data injection attack against power systems

    Int. J. Electr. Power Energy Syst.

    (2019)
  • YiN. et al.

    Task optimization and scheduling of distributed cyber–physical system based on improved ant colony algorithm

    Future Gener. Comput. Syst.-Int. J. Esci.

    (2020)
  • LiangG. et al.

    A review of false data injection attacks against modern power systems

    IEEE Trans. Smart Grid

    (2017)
  • PessimP.S.P. et al.

    State-feedback control for cyber-physical LPV systems under DoS attacks

    IEEE Control Syst. Lett.

    (2021)
  • HumayedA. et al.

    Cyber-physical systems security - A survey

    IEEE Internet Things J.

    (2017)
  • YeD. et al.

    Summation detector for false data-injection attack in cyber-physical systems

    IEEE Trans. Cybern.

    (2020)
  • PengC. et al.

    A survey on security communication and control for smart grids under malicious cyber attacks

    IEEE Trans. Syst. Man Cybern.-Syst.

    (2019)
  • TuH. et al.

    A hybrid cyber attack model for cyber-physical power systems

    IEEE Access

    (2020)
  • LiangG.Q. et al.

    The 2015 Ukraine blackout: Implications for false data injection attacks

    IEEE Trans. Power Syst.

    (2017)
  • TangY. et al.

    Challenge and evolution of cyber attacks in cyber physical power system

  • YangQ. et al.

    On false data-injection attacks against power system state estimation: Modeling and countermeasures

    IEEE Trans. Parallel Distrib. Syst.

    (2014)
  • DengR.L. et al.

    False data injection attacks against state estimation in power distribution systems

    IEEE Trans. Smart Grid

    (2019)
  • XuJ. et al.

    Insulator self-explosion defect detection based on hierarchical multi-task deep learning

    Diangong Jishu Xuebao/Trans. China Electrotech. Soc.

    (2021)
  • YanL. et al.

    Transmission lines modeling method based on fractional order calculus theory

    Diangong Jishu Xuebao/Trans. China Electrotech. Soc.

    (2014)
  • Cited by (41)

    • An improved GraphSAGE to detect power system anomaly based on time-neighbor feature

      2023, Energy Reports
      Citation Excerpt :

      Chao Pei et al. [4] proposed a weighted least squares estimator with an additional Kalman filter, and it can effectively detect FDIA as shown in the experiment results. As for Game theory, [5–7] firstly models data attack and defender sensors from different perspectives, using the game between the defender and the attacker to obtain the best attack injection detection model, and finally, across to the simulation results to prove the effectiveness of the model. Compared with other references, [8] proposed an attack-defense strategy using the theory of optimal economic dispatch, which considers and simulates the change of system state in meeting attacks.

    View all citing articles on Scopus
    View full text